Second Life: Third Party Viewer Meeting (video)
Current RC viewers are all having elevated crash rates. We’ll see new releases of them with various fixes as the Lindens try to get the crash rates down to what we have with the current default viewer, which dropped the crash rate significantly.
So, we very likely won’t see a viewer promoted to main status this coming week 35.
There is a new maintenance viewer coming. It should appear this coming week, but may be not. It could get hung in QA.
Oz Linden is expecting to get the 64-bit version of the viewer out as an RC soon. Oz says it is coming along and they are spending more time on it. There are some ‘corners’ to clean out according to Oz.
The extended skeleton viewer, aka Project Bento, updated in week 34 and is expected to be the last project version. The plan is to go to RC status real soon. From what is being said I would say week 36 or 37.
The viewer’s voice code is being worked on. Oz does not expect a project version to appear too soon as glitches make an ETA for the move from ADITI to Agni servers unpredictable. The Lab and Vivox are meeting next week (35).
On the servers a round of source code clean up changes are in progress. Errors thrown in the render pipeline are getting fixes. The changes will be added to one of the RC channels.
There is a situation now where new accounts don’t show up on Aditi as they should. The Lab is changing how they manage new accounts. They are also adding security to how accounts are stored. That fix is on the way to all the Lab properties. But, for now new people wanting to login into have to Aditi ask support to add them to Aditi’s accounts.
The Aditi password confusion will be fixed so a password change in Agni will immediately take effect on all grids. However, inventory updates will still occur nightly.
Cuing Region Entry
Kitty Barrnett asked about a cue for busy regions to eliminate the use of entry blitz features in HUD’s or built into viewers. People start a HUD to get into a region then go do something in RL. While gone the avatar gets into the crowded region and ends up standing around AFK until the user come back.
The Lab has talked about it internally. But decided it is too problematic. Consider. Does a waiting avatar get a tp request? If so, for how long does the tp pend? What happens to it if the user logs off? Does it block the cue line until something is done with the request? Are others to be blocked by the cue and if so for how long? What happens if the user initiating the request to get in forgets about it? If the user logouts and latter logs back in are they still in the cue?
You can see while answerable the questions get messy.
Oz points out that requests for entry do not place much of a load on the target server. So, the cost of hammering the destination servers is very low. The heavy lifting happens in the departure region. All the scripts and data, the big item, have to be packed up and shipped to the destination server. If the destination servers is full it gets the request, says no, and drops all the packed data all in just a few CPU cycles. If the destination server accepts the entry request then the work starts.
The Lindens are not happy with user experience of having to wait to get into a region. They are open to ideas for how a cue might work. But, Oz prefers to fix regions to hold more people than add a cue. From time to time they bring the idea up and think about it. How full regions handle entry requests is on the list of High Value changes. But, so far they are not ready to try anything.
Ed Merryman, of Firestorm Support, says the entry spammers crash his viewer. On the other hand Whirly says they work well. !?!
An abuse report system that supplies the category list of reportable abuses is being built. So, the idea is then all category lists in all viewers are the same. So, this will be a new server capability to provide the list.
Sending in reports with unsupported categories messes up SL support. That slows things down.
AMD/ATI Video Problem
Hurana Ugajin says they are working with AMD to fix video tearing in the AMD 400 series drivers. We can hope. But, that visual problem has been documented and code for an easy fix supplied to AMD over a year ago and it hasn’t made it out yet. So, I wouldn’t hold my breath.
30:00± – Discussion of voice problems and how servers and voice work together. Oz gets into what they are doing next with voice.
39:00± – Discussion about allocating region resources by parcel. Technically way difficult and so not likely to happen. But, an interesting discussion.
Remainder of the meeting is suggestions and discussion of what might be done to improve this and that.